The Arbitrage That Wasn’t: Why Offshore Savings Rarely Survive Contact with Reality
The forty per cent was never available to be saved.
The Slide That Always Shows Forty Per Cent
Every offshore business case seems to arrive at the same slide. A blended onshore day rate on the left, a blended offshore rate on the right, the ratio between them, and a saving in the high thirties or low forties — a percentage and a three-year cash figure large enough that the discussion effectively ends there. The board nods, the deal is signed, and procurement is congratulated on the rate it has secured per developer-day in Bangalore or Chennai. The programme is then instructed to realise a number that, at the moment of sign-off, has almost no relationship to anything that will actually happen.
Turn the clock forward eighteen or twenty months and the same organisation is often quietly puzzled. The invoices are indeed lower per day. The savings line in the management accounts is not — or not by anything close to the promised figure. Somewhere between the slide and the ledger most of the arbitrage has gone missing, and nobody can quite point to where it went.
I want to argue that this is neither an execution failure nor bad luck. The forty per cent was never available to be saved. It was a labour-rate differential mistaken for a cost saving, and the two are not the same thing.
What a day rate does not price
Let me concede the strongest part of the opposing case first, because sceptics tend to lose this argument by overreaching. The wage gap is real and it is durable. An experienced developer in a tier-one Indian firm costs a fraction of the equivalent in London or Frankfurt, the gap runs to five or six times at the point of hire, and nothing about the next decade obviously closes it. The mistake is not in believing the gap exists. It is in believing you can bank it at the day rate.
A day rate prices an hour of someone’s time in another location. It does not price the work you must do to make that hour useful — and that work is substantial, largely onshore, and almost entirely absent from the comparison slide.
Consider what now has to happen. A requirement that a co-located team would have settled in a corridor in ninety seconds must instead be written down, because the person who will build it is five and a half hours ahead and will read it the next morning while your office sleeps. The specification that was “good enough” for people who shared your assumptions is no longer good enough for people who do not. So a layer of onshore analysts and architects is retained — the shadow team that no business case ever seems to budget for — whose whole job is to convert tacit knowledge into explicit instruction and to reconcile what came back overnight with what was actually meant. Add the defects that escape across the time-zone seam and surface a day later rather than an hour later; add the travel, the dual-running through transition, the network and security work; and add the single most under-modelled figure in the whole exercise — attrition.
In the Indian metros just now, annual attrition of twenty-five to thirty per cent is not a crisis, it is the weather. It means the team that finally understands your business at the end of year one is substantially not the team you begin year two with. Knowledge does not accumulate the way it does in a stable co-located group; it leaks, and it has to be continuously re-poured. That re-pouring is a cost, and you pay it, even though the vendor carries the headcount.
Put a composite but realistic shape on it. Take a hundred-person onshore capability at a blended six hundred a day, moving to a blended offshore engagement rate near three-sixty once the vendor’s own management and margin are loaded in — the headline forty per cent that lands on the slide. Now retain fifteen to twenty people onshore to specify and coordinate; carry twenty to thirty per cent more effort per unit of output through the ramp and the rework; and absorb a quarter of the offshore team turning over every year. What reaches the accounts looks nothing like the slide.
| Cost element | On the business-case slide | In the second-year accounts |
|---|---|---|
| Headline day-rate arbitrage | ~40% | ~40% (unchanged) |
| Retained onshore coordination | Not shown | 15–20% of capability kept onshore |
| Productivity ramp and rework | Not shown | 20–30% more effort per unit early on |
| Attrition and re-ramp | Not shown | ~25–30% annual churn, absorbed continuously |
| Realised saving | ~40% | ~12–18%, arriving in year two |
None of these are exotic costs. They are the ordinary, predictable friction of doing complex knowledge work across distance, language, and time zone. They are missing from the slide not because they are unknowable but because they are inconvenient — and because of who builds the slide.
Why the illusion survives contact with evidence
The comparison persists because it is owned by the function best equipped to measure the one number that is easy and worst placed to measure the ones that are hard. Procurement can establish a day rate to the penny and hold a vendor to it. It cannot see the corridor conversation that no longer happens, the analyst quietly retained, or the release that slips because a clarification waited for morning in two hemispheres. What can be measured precisely drives out what matters more, and the day rate carries the enormous rhetorical advantage of being a single, hard, defensible figure.
There is a quieter incentive at work too. A business case built on forty per cent gets approved; a business case built on the honest fifteen often does not clear the hurdle rate. So the larger number is the one that gets written, and everyone downstream inherits a target that was never real. The programme is then judged against the fiction rather than against what it actually delivered — which is how an initiative that banks a perfectly respectable fifteen per cent ends up recorded as a disappointment.
A durable wage gap is an opportunity, not a saving. Whether any of it reaches the ledger is decided onshore — in how well the work is specified, how much intelligent-client capability is retained, and whether anyone is measuring the cost of delivered output rather than the cost of a day.
The strongest objection, and what it misses
The most serious counter-argument deserves stating at full strength. It runs like this: the arbitrage is structural, not transitional. The wage gap is so large and so durable that even a heavily discounted realisation is worth having; coordination overhead and attrition are teething problems that maturity solves; the leading vendors are assessed at the highest process-maturity levels and improve year on year; therefore the saving is real and merely deferred, not illusory.
Most of that is correct, and it is why offshoring is not going away. But it turns on one move that does the damage — the phrase “teething problems.” Some of the overhead genuinely is transitional and does fall away: the first-year ramp, the initial specification debt, the early defect rate. A large part of it is not transitional at all. The cost of converting tacit knowledge into explicit instruction, of coordinating across a time zone, of re-pouring knowledge through a churning team, is the permanent structural cost of distance. Calling it teething flatters the model and licenses the business case to assume it away.
And here is the sharper point the objection concedes without noticing: the vendor’s rising maturity accrues, by default, to the vendor’s margin. Process improvement on their side lowers their cost of delivery, not your price, unless your contract and your operating model are built to claw it back. The structural gap is real; whether any of it reaches your ledger is an entirely separate question, and it is settled on your side of the ocean.
“The forty per cent is a property of a spreadsheet. The fifteen per cent is a property of an operating model.”
What the organisations that actually bank it do differently
The firms I have watched capture a genuine, durable saving are not the ones that negotiated the hardest rate. They are the ones that stopped treating offshoring as a procurement event and started treating it as an operating-model change — and three habits separate them.
- They invest onshore in exactly the capability the naive case tries to cut. The retained intelligent client — the people who can specify precisely, decompose work for distance, and judge what comes back — is not overhead to be minimised. It is the mechanism through which the wage gap is converted into a saving, and starving it is how organisations turn a seventy per cent rate advantage into a fifteen per cent realised one and then wonder why.
- They measure the cost of delivered output, not the cost of a day. A metric of cost-per-developer-day rewards the very behaviour that destroys value: buying cheap hours regardless of how many it takes to produce a working outcome. Cost per function delivered, or per stable release, tells the truth the day rate hides.
- They write the business case around the number they can keep. Planning to fifteen per cent is not defeatism; it is the difference between a programme that beats its target and one set up to fail against a fantasy. It also changes the sequencing decisions — what to move first, what to hold close, and when a captive centre or a build-operate-transfer arrangement earns its extra cost by protecting knowledge that pure labour arbitrage would let leak away.
The wage gap will still be there in five years, and the organisations that learn to work across distance will have built something their rate-shopping competitors have not: a repeatable capability rather than a line on a slide. The arbitrage is real. It is simply not for sale at the day rate — and the sooner a business case admits that, the more of it the organisation will actually keep.